Traumatised Refugees and Anxiety Traits: Debilitated,Low-income and Unaccompanied During and After Displacement

Abstract

Refugees learned anxiety and trauma from war environments which develops propensity for post-traumatic stress disorder (PTSD). PTSD symptomatology is aggravated when hosting countries have limited knowledge of screening and intervention in received refugees. To add, displaced individuals suffer from pre-existing physical illnesses and other socio-familiar conditions that should be considered in the relationship of likelihood of PTSD and anxiety behaviours. A total of 77 Ukrainian refugees were assessed with the Ukrainian version of PCL-5 instrument regarding PTSD. Respondents answered, in Portugal as hosting country, to 4 symptomatology structure of the scale 12 months after the displacement caused by the Russo-Ukraine War. The survey PCL-5 addressed intrusive memories, avoidance and anxiety behaviour, cognitive and mood alterations and hyper-reactivity. Sociodemographic variables and answering to the PTSD scale generated a training dataset that enabled the development of machine learning (ML) for refugees who are at elevated risk of developing anxiety-related disorders and PTSD. Findings indicated that refugees with higher levels of trauma-related symptoms were identified with pre-existing diseases (musculoskeletal, cardiac and pulmonary), low income (from Ukraine) and moving unaccompanied during the displacement from Ukraine to Portugal. Specifically addressing anxiety, the intrusive memories were the most observed among refugees with the aforementioned characteristics. Intrusive memories are anxious behaviours identified through re-experiencing memories and negative emotions towards traumatic events that individuals were exposed to. Intrusive memories in Ukrainian refugees were greatly associated with avoidance and reactivity occurring on a daily basis. The ML model depicted the refugees at risk for PTSD and allowed for timely and targeted interventions with refugees arriving in the different hosting countries. Results outline the urgency of reducing the environmental stressors in hosting countries, such as fast housing placement and rapid mental health screening that contribute to mitigating existing anxious behaviours as well as the development of new anxiety triggers in the post-migration period.

Keywords

Post-traumatic disorder anxiety machine learning low-income physical illnesses

Introduction

The experience of traumatic events precipitated by armed conflict, compounded by the sudden and often disorienting relocation to host nations, substantially heightens the psychological vulnerability of refugees, rendering them particularly prone to the onset of Post-traumatic Stress Disorder (PTSD), especially within the initial weeks post-displacement. The specific characteristics and intensity of the traumatic exposures critically inform individuals’ affective responsiveness and their predisposition to anticipatory distress when confronted with novel or potentially adverse circumstances (Barbano et al., 2021). The types of losses associated with war significantly increase the likelihood that symptoms of stress and anxiety will develop into severe PTSD when refugees are not monitored through psychological assessments for trauma in their receiving territories. Nevertheless, it is essential to delineate the symptomatological profile of this vulnerable population in order to ensure the accurate differential diagnosis of PTSD and other predominantly anxiety-related psychopathologies. According to DSM-V (2022), PTSD is characterised by behaviour such as intrusive memories, avoidance, arousal and hyper-reactivity and alterations in cognition and mood.

Inherent to the four dimensions of symptomatology, anxiety is very present. Marked hypervigilance is frequently observed, accompanied by a pervasive sense of impending threat (American Psychiatric Association, 2022). These disruptive behaviours are associated with longer or shorter exposure to traumatic events. Accurately discerning the particular groups among refugees and asylum seekers presently manifesting distinct patterns remains a matter of critical urgency to understand the elevated risk of subsequently developing anxiety-related disorders or PTSD. It is crucial to determine, in advance, which refugee populations and asylum seekers have or are at heightened risk for the emergence of anxiety-spectrum psychopathologies or PTSD. To prevent this, early intervention should begin within the first two weeks in host countries by positioning refugees in alignment with established normative frameworks for mental health, independent of cultural or geographical considerations.

Persons with a pre-existing history of mental health disorders and/or sustained exposure to traumatic events are at an elevated risk of developing PTSD when subjected to the stressor of involuntary displacement. Regarding other medical conditions—excluding mental health disorders—we found significant evidence correlating musculoskeletal diseases with anxiety and a tendency to overlap with PTSD, particularly across the four dimensions of symptomatology: intrusive memories, avoidance, hyper-reactivity and cognitive and affective disturbances (Barbano et al., 2021; Bijker et al., 2020; El Sount et al., 2019; Soares & Grossi, 2009). Findings addressed the insufficient investigation about pain and these types of diseases in refugees and with variation in symptomatology for anxiety depending on the living conditions, age and gender (El Sount et al., 2019; Koenen et al., 2017; Teodorescu et al., 2015). In our study, musculoskeletal diseases were prominently reported as pre-existing conditions among Ukrainian refugees. However, there is limited and recent evidence on the prevalence of this relationship between pre-existing physical illnesses and anxiety disorders in populations exposed to traumatic events and war-related conflict (Shalimova et al., 2023).

Other diseases were declared by refugees, such as cardiac and pulmonary chronic cases, which may also be correlated to poor coping strategies during the displacement event after the war blast (Sommer et al., 2019). Pain, mainly from musculoskeletal illness, appears highly associated with PTSD and considering measurement for PTSD in short or longer periods after the exposure to a traumatic event (Koenen et al., 2017; Teodorescu et al., 2015). From this association of physical condition and mental symptomatology of anxiety, avoidance behaviour—one of four PTSD dimensions according to DSM-V—would be prevalent in refugees’ populations (Nissen et al., 2022; Sommer et al., 2019).

This study devised a predictive model employing R, Python and the random forests regression technique to accurately identify refugees at heightened risk, increased susceptibility to the manifestation of anxiety-related disorders and PTSD according to physical pre-existent illness, low income and solo immigration. This model will allow for timely and targeted interventions with refugees arriving into the hosting countries.

Method

Sample

In the present study, we analysed a dataset consisting of responses from 77 adult Ukrainian refugees (M = 43.7 years; SD = 16.16), all of whom underwent clinical interviews and completed the Ukrainian version of the PTSD checklist for DSM-5 (PCL-5). The participants were Ukrainian nationals originating from various cities, predominantly Mariupol, Kharkiv and Kyiv. The individuals arrived in Portugal subsequent to the onset of the war in February 2022 and were subsequently integrated into a reception programme within one of the interfamily units located in the Lisbon metropolitan area. These units offer comprehensive support, encompassing basic necessities such as sustenance, essential healthcare and provisional lodging, with the possibility of transitioning to permanent housing outside the designated family intervention units. The programme ensures the reunification of families, even when members arrive in the host country at different intervals.

In terms of employment, only a minority were employed—4 individuals—while 3 were retirees and 70 were without employment at the time of the survey. Despite a significant proportion (90%) being unemployed, attributed to their recent refugee status, an impressive 79% of participants held higher education qualifications, including bachelor’s and master’s degrees, while the remainder possessed secondary-level qualifications. Regarding income, the average amount was considerably lower than the Portuguese average, with a mean of 374 euros (SD = 377.41), in stark contrast to the typical earnings within Portugal. All respondents were proficient in Ukrainian, 48% were fluent in Russian and 35% were conversant in English. According to the sociodemographic data, 61% of the sample reported chronic physical conditions, predominantly affecting the cardiovascular, respiratory and musculoskeletal systems. None of the respondents had been diagnosed with any mental health disorders.

The inclusion criteria for participation in this study were prior residence in Ukraine, refugee-approved status in Portugal as a result of the ongoing invasion of Ukraine, adulthood, a minimum residency period of 4 weeks in Portugal and the absence of psychological or psychiatric conditions, as well as the lack of engagement in any therapy (of psychological origin) programmes.

Instrument

The PTSD checklist for DSM-5 (PCL-5) is a standardised self-administered instrument designed to evaluate post-traumatic symptomatology based on the 20 diagnostic criteria outlined in the DSM-5. The items are systematically categorised into four symptom clusters—Intrusion (Cluster B), Avoidance (Cluster C), Negative Alterations in Cognition and Mood (Cluster D) and Alterations in Arousal and Reactivity (Cluster E). Respondents assess each item on a 5-point Likert scale ranging from 0 (Not at all) to 4 (Extremely), with elevated scores reflecting an increased severity of post-traumatic stress manifestations.

Criterion B: Intrusive Symptoms

Criterion B evaluates the presence of intrusive symptoms, characterised by recurrent, distressing and involuntary recollections of the traumatic experience. This encompasses persistent, unsettling dreams related to the event, as well as flashbacks, during which the individual feels as though the traumatic event is occurring once more. Furthermore, this criterion examines the individual’s emotional and physiological responses to trauma reminders, which may include heightened distress or physiological reactions, such as elevated heart rate and perspiration.

Criterion C: Avoidance

Criterion C focuses on avoidance behaviours, where individuals actively seek to distance themselves from memories, thoughts, or emotions associated with the traumatic event. This pattern of avoidance may encompass external stimuli—including specific individuals, locations, conversations or activities—that serve as reminders of the traumatic event. As a core feature of PTSD, avoidance behaviours can profoundly disrupt daily functioning and markedly diminish overall quality of life.

Criterion D: Negative Alterations in Cognition and Mood

Criterion D assesses detrimental shifts in cognition and mood observed after trauma. These alterations encompass a range of cognitive and emotional disruptions, including an impaired ability to recall critical details of the traumatic experience, pervasive and maladaptive negative beliefs about oneself or the external world, as well as distorted feelings of self-blame or culpability directed towards others. Additional symptoms within this domain include persistent negative emotional states, such as fear, anger, guilt or shame, a marked disinterest in activities once found pleasurable, a pervasive sense of detachment from others and an impaired capacity to experience positive emotions.

Criterion E: Alterations in Arousal and Reactivity

Criterion E explores alterations in arousal and reactivity following exposure to trauma. Symptoms within this category encompass irritability or increased aggressiveness, engagement in reckless or self-destructive behaviours, hypervigilance (characterised by heightened alertness and defensive vigilance), exaggerated startle responses, difficulties with concentration and significant sleep disturbances, including problems with both the initiation and maintenance of sleep. The PCL-5 assessment utilises a 5-point Likert scale, with scores ranging from 0 (Not at all) to 4 (Extremely). The total score can range from 0 to 80, with higher scores indicating more severe PTSD symptoms. All clusters are indicative of anxiety-related traits.

The established cut-off score for the total scale ranges between 33 and 38, as outlined in the original reference of the instrument. The maximum possible score on the PCL-5 is 80. This tool is employed both in clinical practice and research contexts to assess the presence and intensity of PTSD symptoms. A Ukrainian version of the PCL-5 was administered in 2023 (National Centre for PTSD) and was found to be suitable for evaluating PTSD in Ukrainian refugees residing in Portugal, with a reliability coefficient (α = 0.83) (Figueiredo et al., 2024).

Procedure and Ethics

The training dataset was derived through the administration of two psychometric instruments to a cohort of Ukrainian refugees, who had been mobilised from conflict zones beginning in late February 2022. That training base is part of a current project authorised by the Editorial Board of Ethics in Psychology from (University Research Center in Psychology [CUIP]).

The cohort of this study is divided into three distinct groups based on their respective arrival periods in Portugal: March, April and May of 2022. Upon receiving approval from the Ethics Committee, the research protocol was formally established. The study was sanctioned for academic and scientific objectives, and the participation rate was overwhelmingly high. In the second phase, the application of the questionnaires assured by the research team in conjunction with a social worker and a psychologist from the collaborating institution was accomplished. The psychologist exhibited advanced proficiency in Ukrainian, English and Portuguese, which facilitated their collaboration in presenting the study and clarifying respondents’ doubts.

Specific sessions were organised to facilitate brief groups in completing the questionnaire at their own pace, with support provided by the research team. These sessions were conducted in February 2023, in a room provided by the host institution. To ensure confidentiality and ethical principles inherent to research protocol, appropriate procedures were implemented both during and after the administration of the questionnaires.

During the data processing phase, it was essential to guarantee the accurate translation of responses provided in Ukrainian. A psychologist was enrolled during the sessions to assist with this task while maintaining the anonymity of the participants. The database was thoroughly completed and reviewed in preparation for the predictive evaluation study utilising machine learning (ML) techniques.

Data Analysis

We ran a computational analysis in R software, version R 3.6.1 and RStudio (Version 3.5.3). A distinctive feature of this programme is its capacity to predict temporal sequences with both precision and efficiency, regardless of the data corpus. Previously, sociodemographic characterisation was based on information gathered from the first questionnaire. The data were instrumental in assessing the impact of various interacting factors on participants’ responses to the PCL-5. Therefore, in addition to utilising the R software, the analysis and methodological approach were further enhanced by the inclusion of the Python package, a more accessible programming language that facilitates ease of reading and replication by researchers in the social sciences field.

Subsequently, an ML model was developed to predict PTSD, taking into account the external factors that interact with the four clusters of the PCL-5. By external factors, we refer to independent variables—such as low income, solo immigration and existing physical illnesses (unrelated to diagnosed mental health disorders)—which are peripheral but integral to the overall diagnostic assessment framework.

Results

A total of 39 of 77 Ukrainian refugees met PTSD according to PCL-5 results (Figueiredo et al., 2024), with the minimum score established at 33 points. Less than 2 points per item determined exclusion of participants for PTSD. In a previous study with the sample, we found that monthly income (in Ukraine) and household (and civil status) showed strong variability (household: Z = –5.779, p < .05; marital condition: Z = –5.713, p < .05; monthly income: Z = –5.467, p < .05) across the clusters, especially found for negative alteration in cognition and mood. These preliminary results are followed by the present model learning depicted for a profound analysis of PTSD, considering the factors in reference (income and household, as well as mobilisation alone or with relatives).

On the other hand, through discriminant analysis using the Mann-Whitney U test, only the factor ‘illness’ (physical) revealed significant discrimination, with a high effect size, for the symptomatology structured in the four-factorial model of PTSD (W = .988; Z = 6.378; df = 1.71; p = .014). More anxiety associated with cognitive alterations and reactivity behaviour was verified more in respondents with self-reported diseases (musculoskeletal, cardiac and pulmonary issues). To add, refugees that had moved alone Ukraine directly to Portugal, with no camp refugees’ experiences in the middle of the displacement process, suffered more from cognitive and mood alterations compared to their peers that moved at the same time but were accompanied by relatives. It is important to point out the fact that 74% of participants self-reported a solo displacement (moving to another country with no relatives because they have no one since their origin or because of the death of relatives living in household context, caused by war).

Feature Importance Analysis: Random Forests Regression

The ML conducted in this study was motivated by limitations observed in previous research for PTSD diagnosis with the same sample (Figueiredo et al., 2024). The PCL-5 structure was revisited towards the original four-factorial model of Weathers et al. (2013). In fact, in that previous study, we found through an exploratory factorial analysis a non-desirable and unexpected dispersion of items identified as being part of Cluster E (cognitive and mood changes behaviour). One suggestion was made on that evidence: to compute items from Clusters D and E into one single cluster. Here, the ML model allowed to refine the relevance of clusters according to specific variables. In Cluster B, referring to intrusive memories, the loadings were higher for the factorial structure. Thus, our current findings overlap that limitation about using all clusters for different conditions of refugees. To add, the same Cluster B was loading also in the best way for the ill, low-income and ‘alone’ participants.

Findings with the ML model, specifically using Radom Forest, showed a high correlation (p < .05) of skeletal and cardiac diseases with high levels in PTSD assessment. Individuals with more diseases reflected more punctuations among the four subscales of PCL-5. Random Forests developed a model where low-income status associated with poor work situations in adult refugees was strongly associated with high levels of avoidance behaviour and hyper-reactivity (high arousal behaviours).

Receiver Operating Characteristic: AUC

The area under the curve (AUC) detected that, for Cluster B (intrusive memories), low income and physical illness were higher predictors when compared to other variables. Refugees re-experience more war memories when they have low income and physical diseases, specifically musculoskeletal, cardiac and pulmonary. Household income, using the salary reference from Ukraine, significantly influences the model’s performance. In Cluster B, the AUC (> 0.50) was positive, yet the curve’s coordinates revealed suboptimal results, with 50% of the predictions being false positives. The cutoff point was set at the lowest salary level (275 euros per month, with the minimum reported salary being 15 euros). A similar outcome was found in Cluster E (hyper-reactivity), where an AUC > 0.50 was observed for the low salary threshold (275 euros per month), but the specificity indices were unacceptable, as over 50% of predictions were false positives. Conversely, in Cluster C, the AUC reached its highest value (= 0.92), where the highest income (837 euros) resulted in fewer than 1% false positives. This suggests that nearly all the responses form a near-perfect model for predicting the correlation between higher income levels and an increased occurrence of intrusive memories. Other variable emerged with explaining variance (AUC = 0.84) in the model regarding the re-experiencing events: solo immigration differed in symptomatology compared to refugees accompanied with families. Specificity values proved to be weak, with only 28% being false positives, with values = 1 regarding sensitivity (true positives). Divorced and widowed subjects presented the most problems with intrusive war-related memories, in contrast to single and married subjects (accompanied with relatives).

Although the AUC = 0.50 for Clusters D (Negative Cognitive and Emotional Changes) and E (Alterations in Reactivity and Arousal) was relatively modest, it still met acceptable standards. However, when examining specificity with a cutoff point based on marital status (divorced or widowed), the results indicated a 43% false positive rate—almost half of the cases. This suggests that the impact of symptoms in these clusters is less pronounced when considering the marital status of Ukrainian refugees in Portugal. More specifically, refugees who migrate and settle alone are more likely to experience higher levels of traumatic memories and heightened reactivity.

In relation to household dynamics during migration, Cluster B showed a satisfactory AUC (= 0.50), with cutoff points of 2.5 and 3, representing the number of household members. Though minimal, the specificity showed a relatively low false positive rate (28%), indicating that household structure plays an important role in predicting the relationship explored in this cluster.

Cluster D exhibited a more robust predictive model (AUC = 0.84), demonstrating that changes in cognition and mood were influenced by household composition. Larger household sizes were associated with an increase in true positive responses, with only 29% false positives. A similar trend was observed in Cluster E (hyper-reactivity), although the AUC was slightly lower (0.70).

The curve analysis is presented in Figure 1.

Figure 1.

ROC Plot of Intrusive Memories and Cognitive/Mood Alterations.

Considering the factors examined, the pre-existing illness, the income and the household (as well as the marital status related to the household variable to explain solo immigration cases—with no one during the mobility), we also ran a univariate analysis of variance (ANOVA) considering the four clusters of the scale. The answer to each cluster is in the training data set used for the ML developed for this study. The results from ANOVA and the Tukey post-hoc test confirmed that musculoskeletal and cardiac diseases are linked, with a great effect size (η² > .06), to specific clusters of the PTSD scale, which enables predicting anxiety risk for this population. Anxiety is inherent to the clusters identifying intrusive memories (F = 118.09; df = 1.72; p = .028) and avoidance behaviour (F = 34.56; df = 1.73; p = .012) among refugees with physical illness.

For the other factors—income and aggregate composition during mobility—ANOVA did not reveal statistically significant differences or magnitudes that confirmed the impact of those factors on PTSD and anxiety behaviours related to the war conflict and to the displacement.

To summarise, several factors emerged as strong predictors of PTSD, including low monthly income according to Ukrainian standards, migration without familial support, smaller household sizes (where individuals with fewer relatives exhibited increased reactivity and cognitive changes) and the presence of chronic physical conditions, particularly skeletal and cardiac diseases. These variables were most strongly associated with intrusive memories and avoidance behaviours. Both re-experiencing the trauma and avoidance are closely tied to the anxiety caused by heightened vigilance, which occurs daily and significantly disrupts the cognitive processes of affected individuals.

Discussion

Results from the ML model built for this study allowed us to understand how PTSD likelihood in refugees from Ukraine is correlated with the fact of having physical diseases, low income and household (or even moving alone during the displacement process). The model refers mainly to the Cluster B of the four-model of PTSD (APA, 2022), thus addressing the intrusive memories or re-experiencing traumatic events. The constant re-experiencing of events is elicited by the fact that the Russo-Ukrainian War is on course and in an acute phase. Memories might be repetitive considering the information provided by social media and the contact that refugees have with their relatives living in the conflict areas.

Intrusive and repetitive memories refer to events related to the Ukrainian war and to the forced migration episode. Avoidance behaviour was also identified with expression among the examinees. The avoidance generates negative emotions that re-traumatise participants through the repetitive memories. Both clusters of intrusive memories and avoidance are related. Note also that hosting conditions may act as a facilitating factor of that symptomatology when housing, health services access, communication and education are not adequately provided, facing the patterns and routines that these adults had back in Ukraine. Financial security is another problem eliciting anxious behaviours and operating as a risk factor for more intrusive memories linked to the war. Routines disruption followed by no occupational activities are other risk factors for intrusive memories and negative arousal in refugees.

These consequences from risk factors associated with intrusive memories in refugees were studied in recent research (Badawi et al., 2022; Höhne et al., 2023; Nickerson et al., 2023), mainly regarding the uncertainty effects for intrusive fear and lack of patience. Regarding avoidance, on the other hand, previous studies determined avoidance related to anxiety and poor self-regulation among refugees from Syria (Doolan et al., 2017). The background of refugees (like the country and war of origin) makes the difference for PTSD and related symptoms (Almoussa & Mattei, 2023; Melese et al., 2024). However, in Doolan et al.’s (2017) study, avoidance symptoms were already linked to memories of war. Nevertheless, very few studies comprehensively address the significance of this background analysis.

Regarding the refugees’ background, other characteristics should be examined in the burden of individuals subjected to involuntary migration. Thus, the results obtained using decision trees (feature importance) indicated that refugees who arrived in Portugal within one month and had pre-existing heart conditions (such as cardiac problems and hypertension) were particularly susceptible to developing PTSD. Specific physical illnesses (musculoskeletal, cardiac and pulmonary) demonstrated to have predictive value for PTSD with a tendency to intensify anxiety symptoms. Intrusive memories and avoidance behaviour are intrinsically related to the anxiety symptoms and likelihood of anxiety panic and other related disorders. In specific studies of Piotrowicz et al. (2022) and Shalimova et al. (2023), the heart diseases were also observed in Ukrainian refugees. An affirmative connection was further observed for PTSD. Lolk et al. (2016) and Al-Rousan et al. (2023) found a direct correlation between PTSD and cardiac as well as pulmonary diseases in immigrants and refugees. Both are displaced populations for distinct reasons in origin, but immigrants share the same condition of displacement and stressful host conditions until settlement.

In our sample we also found pulmonary conditions, among cardiac and musculoskeletal illnesses. All individuals suffering from those diseases had PTSD after our diagnosis conducted through PCL-5. This data are supported by studies of Shalimova et al. (2023). Moran et al. (2023), Rosenthal et al. (2022) and van den Berk et al. (2022) represent the inverse correlation in their studies: PTSD related to forced mobilisation in refugees predicts the development of heart diseases. To our knowledge, literature mostly refers to that correlation and keeps understudied the pre-existing disease of refugees in correlation with the escalating development of PTSD. Considering results from previous existent research and adding findings of our current study, we may conclude that pre-existing heart diseases became more prominent with the appearance of PTSD.

About the income effect for PTSD and specific anxiety features, we found that low-income refugees demonstrated higher anxious behaviour expressed by re-experiencing events through memories of war. The re-experience might be associated with the post-migration period where anxiety is more developed compared to the period of pre-migration (Kartal & Kiropoulos, 2016). Anxious behaviours related to intrusive and negative memories are poorly explored in research. However, anxiety and PTSD in refugees are well documented, particularly in studies addressing income disparities among refugee populations. Monthly incomes refer to Ukrainian salaries of participants that were terminated since the forced displacement started. Low-income suggests absence or low opportunity of savings (Choy et al., 2021). This constrained the living conditions in the host country. Ukraine is well known as a poor country in terms of average income and cost of life. Portugal has a high cost of living but a low average income (however much higher than the Ukrainian reference). Hosting refugees with low financial resources in Western countries intensifies anxious acculturation, which is perceived as a stressful factor contributing to social anxiety (Garbade et al., 2023; Usama et al., 2021).

The insights from our predictive model developed in ML help in forming hypotheses and guiding further research. For example, understanding that lower income (based on the reference of income for Ukraine) is heavily associated with higher PTSD symptoms may serve as a basis for the development of targeted support programmes for refugees with lower socio-economic status, particularly those who are not employed in Portugal and are likely to have limited financial resources; therefore, they struggle with expenses of health services in Portugal. This is the baseline to understand where the mitigation of post-migration difficulties in refugees begins. As a computed equation, the trained Random Forest model revealed several critical results, and a range of factors have been identified as contributing to the heightened risk of PTSD among Ukrainian refugees. Notably, individuals with low monthly incomes (prior to displacement) and those with pre-existing health conditions are categorised as belonging to high-risk groups. These factors, in conjunction with the stressors associated with displacement, significantly amplify the vulnerability to post-traumatic stress.

Also, household (involving the marital status and the co-migration of relatives) emerged as a variable of substantial predictive importance within the Random Forest regression model for the assessment of PTSD. Additionally, the structural configuration of familial aggregates, specifically the number of relatives present during the displacement process, was corroborated as a critical determinant, exhibiting a high predictive value for PTSD as evidenced by the results of the receiver Operating Characteristic analysis. Here, we call ‘solo’ immigration or ‘solo’ displacement for the status moving unaccompanied. Some studies examined the unaccompanied refugees and PTSD, but only referring to children (Carlson et al., 2012; Daniel-Calveras et al., 2022; Höhne et al., 2023; Hornfeck et al., 2022; Huemer et al., 2009).

Adults are neglected so far in research, with probably high consequences for intervention. Our results showed that unaccompanied adults punctuated more in anxiety-related items of PCL-5. They suffer more than their accompanied peers during the post-migration process. Adults revealed more stress and anxiety due to the uncertainty that they are aware of (Badawi et al., 2022; Höhne et al., 2023; Nickerson et al., 2023). The war events are not perceived (and re-experienced) in the same terms as minor children do . As well, adults need to interact with new events that are forcibly stressful such as the network and communication in the new labour market, housing pursuits and social adjustments (Wallin & Ahlström, 2005). More and recent research is need to understand the impact of being an adult refugee in a post-migration period regarding PTSD and anxiety symptomatology, especially for low-income, alone and ill refugees.

Conclusion and Clinical Implications

By employing sophisticated methodologies and analytical techniques, the study pioneered the creation of a highly effective predictive model utilising the Random Forest algorithm to assess PTSD risk among Ukrainian refugees. This extensive approach not only guaranteed the model’s reliability and operational efficiency but also enhanced its interpretability, providing crucial understanding of the underlying factors contributing to PTSD and supporting the development of focused mental health strategies.

The model enables to understand which refugees are at higher risk and developing rapidly PTSD: low income, moving alone into Portugal and having pre-existing illnesses (musculoskeletal, cardiac and pulmonary) showed to be the most prevalent risk factors, mainly for the re-experiencing of negative memories. On the other hand, refugees in second priority for intervention would exhibit the following characteristics: no physical illnesses, high income (received in Ukraine) and being accompanied during forced displacement and in the hosting country.

This model, therefore, presents the potential for replication across diverse refugee nationality cohorts, incorporating a broad spectrum of additional variables. These include, but are not limited to, the duration of residence in host countries, the cities from which displacement originated, the cities within the host country of resettlement, as well as the economic, professional and academic qualifications of the refugees. The inclusion of such multifaceted factors would empower healthcare providers to systematically prioritise interventions, thereby facilitating a more nuanced and contextually informed approach to the healthcare needs of refugees.

The refugee’s transitional situation is opaque for the understanding of professionals that receive the newcomers in a first instance. Thus, the priority would be tailoring intervention and screening according to specific characteristics of groups. Reducing the environmental stressors in hosting countries, such as the shortened periods of the documentation process and fast housing placement, will contribute to mitigating existing anxious behaviours as well as the development of new anxiety triggers that may constrain the PTSD treatment when refugees already have PTSD. However, it should be noted that although the PCL-5 can be used for PTSD screening and provisional diagnosis, the clinical interview should always be present to complete the assessment of PTSD. Pre-migration and post-migration periods are related to different anxiety events. Post-migration challenges can be more pronounced due to the language barrier when refugees communicate in the language of the host country, compounded by the unfamiliarity of both local communities and professionals. This can lead to high levels of mistrust and defiance, as well as the re-experiencing of traumatic war memories. Applying the model built in this study would help to identify emergent anxious groups (suffering from trauma or incurring trauma experience) and save time to diagnose and develop an effective support programme designed for those groups with higher risk evaluated.

Footnotes

Data Availability

Data generated and analysed during this study are included and clearly identified in this article.

Declaration of Conflicting Interests

The authors declared on potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Ethical Approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional Research Committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The Ethics Committee of Universidade Autónoma de Lisboa granted the approval as well as the Foundation for Science and Technology.

Funding

The authors disclosed the following funding for the authorship and/or publication of this article: This work was funded by national funds through FCT—Fundação para a Ciência e a Tecnologia (Foundation for Science and Technology)—as part of the project CIP/UAL—Ref. UIDB/04345/2020 and the Psychology Research Centre (CIP) of Universidade Autónoma de Lisboa/Universidade do Algarve. The English translation of this work was funded by national funds through FCT—Fundação para a Ciência e a Tecnologia—as part of the CIP/UAlg project—Ref. UIDB/04345/2020.

Informed Consent

The consents were delivered, ensuring the anonymous principle of data collection, treatment and dissemination (for academic purposes only). All the participants accorded with the consent statement.

Permission to Reproduce Material from Other Sources

There are no third-party sources used in this study; all the data and figures are original. Authors granted full permission, on request and citation, to the research community to use/replicate this material.

References

American Psychiatric Association (APA). (2022). Diagnostic and statistical manual of mental disorders (5th ed.) American Psychiatric Association.

Almoussa

, & Mattei

(2023). Cardiovascular health and risk factors in African refugees and immigrants in the United States: A narrative review. Ethnicity & Health, 28(3), 399–412. https://doi.org/10.1080/13557858.2022.2052712

Al-Rousan

, AlHeresh

, Saadi

, El-Sabrout

, Young

, Benmarhnia

, & Anagnostaki

, & Zaharia

(2023). “Children don’t think about these things”: Qualitative research on the relationship views and experiences of former unaccompanied refugee children. Journal of Social and Personal Relationships, 40(11), 3464–3484. https://doi.org/10.1177/02654075231175633

Badawi

, Steel

, Harb

, & Mahoney

(2022). Changes in intolerance of uncertainty over the course of treatment predict posttraumatic stress disorder symptoms in an inpatient sample. Clinical Psychology & Psychotherapy, 29(1), 230–239. https://doi.org/10.1002/cpp.2625

Barbano

A. C.

, Tull

M. T.

, Christ

N. M.

, Xie

, Kaminski

, & Wang

(2021). Fear of pain as a predictor of concurrent and downstream PTSD symptoms. Journal of Anxiety Disorders, 82, 102441. https://doi.org/10.1016/j.janxdis.2021.102441

Bijker

, Sleijser-Koehorst

M. L. S.

, Coppieters

M. W.

, Cuijpers

, & Scholten-Peeters

G. G. M.

(2020). Preferred self-administered questionnaires to assess depression, anxiety and somatization in people with musculoskeletal pain—a modified Delphi study. The Journal of Pain, 21(3–4), 409–417. https://doi.org/10.1016/j.jpain.2019.08.006

Carlson

B. E.

, Cacciatore

, & Klimek

(2012). A risk and resilience perspective on unaccompanied refugee minors. Social Work, 57(3), 259–269. https://doi.org/10.1093/sw/sws003

Choy

, Arunachalam

, Gupta

, Taylor

, & Lee

(2021). Systematic review: Acculturation strategies and their impact on the mental health of migrant populations. Public Health in Practice, 2, 100069. https://doi.org/10.1016/j.puhip.2020.100069

Daniel-Calveras

, Baldaquí

, & Baeza

(2022). Mental health of unaccompanied refugee minors in Europe: A systematic review. Child Abuse & Neglect, 133, 105865. https://doi.org/10.1016/j.chiabu.2022.105865

10.

Doolan

E.L.

, Bryant

R.A.

, Liddell

B.J.

, & Nickerson

(2017). The conceptualization of emotion regulation difficulties, and its association with posttraumatic stress symptoms in traumatized refugees. Journal of Anxiety Disorder, 50, 7–14. https://doi.org/10.1016/j.janxdis.2017.04.005

11.

El Sount

C. R. O.

, Windthorst

, Denkinger

, Ziser

, Nikendei

, Kindermann

, & Junne

(2019). Chronic pain in refugees with posttraumatic stress disorder (PTSD): A systematic review on patients’ characteristics and specific interventions. Journal of Psychosomatic Research, 118, 83–97. https://doi.org/10.1016/j.jpsychores.2018.07.014

12.

Figueiredo

Dierks

, & Ferreira

(2024). Mental health screening in refugees’ communities: Ukrainian refugees and their post-traumatic stress disorder specificities. European Journal of Trauma & Dissociation, 8(1), 100382. https://doi.org/10.1016/j.ejtd.2024.100382

13.

Garbade

, Eglinsky

, Kindler

, Rosner

, Sachser

, & Pfeiffer

(2023). Factors affecting the acculturation strategies of unaccompanied refugee minors in Germany. Frontiers in Psychology, 14, 1149437. https://doi.org/10.1016/j.jpsychores.2020.110175

14.

Kartal

, & Kiropoulos

(2016). Effects of acculturative stress on PTSD, depressive, and anxiety symptoms among refugees resettled in Australia and Austria. European Journal of Psychotraumatology, 7(1), 28711. https://doi.org/10.3402/ejpt.v7.28711@zept20.2016.7.issue-s1

15.

Koenen

L. R.

, Icenhour

, Forkmann

, Pasler

, Theysohn

, Forsting

, & Elsenbruch

(2017). Greater fear of visceral pain contributes to differences between visceral and somatic pain in healthy women. Pain, 158(8), 1599–1608. https://doi.org/10.1097/j.pain.0000000000000924

16.

Höhne

, Banaschewski

, Bajbouj

, Böge

, Sukale

, & Kamp-Becker

(2023). Prevalences of mental distress and its associated factors in unaccompanied refugee minors in Germany. European Child & Adolescent Psychiatry, 32(7), 1211–1217. https://doi.org/10.1007/s00787-021-01926-z

17.

Hornfeck

, Sowade

, & Bovenschen

(2022). Effects of the asylum process on the mental health of unaccompanied young refugees—A scoping review. Children and Youth Services Review, 137, 106490. https://doi.org/10.1016/j.childyouth.2022.106490

18.

Huemer

, Karnik

N. S.

, Voelkl-Kernstock

, Granditsch

, Dervic

, Friedrich

M. H.

, & Steiner

(2009). Mental health issues in unaccompanied refugee minors. Child and Adolescent Psychiatry and Mental Health, 3, 1–10. https://doi.org/10.1186/1753-2000-3-13

19.

Moran

J. K.

, Jesuthasan

, Schalinski

, Kurmeyer

, Oertelt-Prigione

, Abels

, & Schouler-Ocak

(2023). Traumatic life events and association with depression, anxiety, and somatization symptoms in female refugees. JAMA Network Open, 6(7), e2324511–e2324511. https://doi.org/10.1001/jamanetworkopen.2023.24511

20.

Lolk

, Byberg

, Carlsson

, & Norredam

(2016). Somatic comorbidity among migrants with posttraumatic stress disorder and depression—a prospective cohort study. BMC Psychiatry, 16, 1–12. https://doi.org/10.1186/s12888-016-1149-2

21.

Melese

, Simegn

, Esubalew

, Limenh

L. W.

, Ayenew

, Chanie

G. S.

, & Gela

Y. Y.

(2024). Symptoms of posttraumatic stress, anxiety, and depression, along with their associated factors, among Eritrean refugees in Dabat town, northwest Ethiopia. BMC Psychology, 12(1), 62. https://doi.org/10.1186/s40359-024-01554-7

22.

Nickerson

, Hoffman

, Keegan

, Kashyap

, Argadianti

, Tricesaria

, & Liddell

(2023). Intolerance of uncertainty, posttraumatic stress, depression, and fears for the future among displaced refugees. Journal of Anxiety Disorders, 94, 102672. https://doi.org/10.1016/j.janxdis.2023.102672

23.

Nissen

, Hynek

K. A.

, Scales

, Hilden

P. K.

, & Straiton

(2022). Chronic pain, mental health and functional impairment in adult refugees from Syria resettled in Norway: A cross-sectional study. BMC Psychiatry, 22(1), 571. https://doi.osrg/10.1186/s12888-022-04200-x

24.

Piotrowicz

, Semeniv

, Kupis

, Ry’s

, Perera

, Gryglewska

, & Gąsowski

(2022). Disease burden in older Ukrainian refugees of war: A synthetic reanalysis of public records data. The Lancet Healthy Longevity, 3(10), e667–e673. https://doi.org/10.1016/s2666-7568(22)00187-8

25.

Rosenthal

, Touyz

R. M.

, & Oparil

(2022). Migrating populations and health: risk factors for cardiovascular disease and metabolic syndrome. Current Hypertension Reports, 24(9), 325–340. https://doi.org/10.1007/s11906-022-01194-5

26.

Shalimova

, Stoenoiu

M. S.

, Cubala

, Burnier

, Persu

, & Narkiewicz

(2023). Association of high blood pressure with post-traumatic stress disorder in Ukrainian women seeking asylum in Poland. Journal of Hypertension, 41(Suppl. 3), e20. https://doi.org/10.1097/01.hjh.0000939068.66076.cb

27.

Soares

J. J.

, & Grossi

(2009). The relationship between levels of self-esteem, clinical variables, anxiety/depression and coping among patients with musculoskeletal pain. Scandinavian Journal of Occupational Therapy, 7(2), 87–95. https://doi.org/10.1080/110381200750018887

28.

Sommer

J. L.

, El-Gabalawy

, & Mota

(2019). Understanding the association between posttraumatic stress disorder characteristics and physical health conditions: A population-based study. Journal of Psychosomatic Research, 126, 109776. https://doi.org/10.1016/j.jpsychores.2019.109776

29.

Teodorescu

D. S.

, Heir

, Siqveland

, Hauff

, Wentzel-Larsen

, & Lien

(2015). Chronic pain in multi-traumatized outpatients with a refugee background resettled in Norway: A cross-sectional study. BMC Psychology, 3, 1–12. https://doi.org/10.1186/s40359-015-0064-5

30.

Usama

E. A.

, Fathi

, Vasileva

, Petermann

, & Reinelt

(2021). Acculturation orientations and mental health when facing post-migration stress: Differences between unaccompanied and accompanied male Middle Eastern refugee adolescents, first-and second-generation immigrant and native peers in Germany. International Journal of Intercultural Relations, 82, 232–246. https://doi.org/10.1016/j.ijintrel.2021.04.002

31.

van den Berk Clark

, Kansara

, Fedorova

, Ju

, Renirie

, Lee

, & Scherrer

J. F.

(2022). How does PTSD treatment affect cardiovascular, diabetes and metabolic disease risk factors and outcomes? A systematic review. Journal of Psychosomatic Research, 157, 110793. https://doi.org/10.1016/j.jpsychores.2022.110793

32.

Wallin

A. M. M.

, & Ahlström

G. I.

(2005). Unaccompanied young adult refugees in Sweden, experiences of their life situation and well-being: A qualitative follow-up study. Ethnicity & Health, 10(2), 129–144. https://doi.org/10.1080/13557850500071145

33.

Weathers

F. W.

, Litz

B. T.

, Keane

T. M.

, Palmieri

P. A.

, Marx

B. P.

, & Schnurr

P. P.

(2013). The PTSD Checklist for DSM-5 (PCL-5). http://www.ptsd.va.gov